Future Made in Australia

Kirsty Waldron moved her family from Rockhampton to Gladstone in the middle of 2023 because a hydrogen plant was coming. She is a fitter and turner by trade, fifteen years of maintenance shutdowns on alumina refineries and LNG trains behind her, and she had read the same announcements everyone in central Queensland had read: billions of dollars of investment, a new export industry, thousands of construction jobs and then hundreds of permanent ones, all of it anchored on the deep-water port and the heavy-industrial land the town already had. She signed a two-year lease at a Gladstone rent, enrolled her younger child in a new school, and paid out of her own pocket for a short course in the handling of high-pressure gases, on the straightforward reasoning that the people qualified first would be the people hired first. Her employer, a mid-sized mechanical contracting firm, took on extra apprentices on the same logic and quoted for shed space it did not yet need.

By late 2025 the plant she moved for was not being built. The proponent had walked away, the Queensland Government had withdrawn a billion dollars of state support from a related Gladstone venture, and a project with an announced capital value of roughly $14 billion had gone from a construction pipeline back to a media release. Kirsty is still working — the LNG shutdowns still run, and central Queensland is not short of maintenance work. Her name and her details are a composite rather than one person's, but the position is not invented: across Gladstone, the Hunter Valley, the Pilbara and the upper Spencer Gulf, workers, contractors, councils and training providers geared themselves to a renewable-hydrogen industry on the strength of announcements, and then watched the proponents of those announcements withdraw one after another through 2024 and 2025. What turns her situation from a private disappointment into a policy question is what the Commonwealth did next. At close to the moment the companies with their own capital at risk were leaving, the Commonwealth legislated a subsidy to pay whoever stays.

That sequence is the subject of this article, and it produces an unusually clean analytical test. Public subsidy is normally defended on the ground that a market is failing — that private investors, left alone, will under-supply something the country needs, either because the benefits spill over to people who do not pay for them or because the risk is of a kind private balance sheets cannot carry. That is a serious argument, and in parts of this policy it is a strong one. But when the commercial proponents of a specific activity, holding their own money, are exiting that activity at speed, the question of whether a subsidy is correcting a market failure or funding something the market has already priced correctly stops being abstract. It becomes a matter of evidence. This article separates the agenda's two halves — critical minerals and hydrogen — because the evidence points in different directions for each, and treating them as one undifferentiated thing is the fastest way to get the judgement wrong.

What the government did

The Future Made in Australia agenda has two layers: a framework Act that sets out how the Commonwealth decides which industries to back, and a second Act that creates the money. The Future Made in Australia Act 2024 established what it calls the National Interest Framework — a statutory test, administered by the Treasury, for deciding whether a sector qualifies for public support. The Framework runs two streams. The first is the net zero transformation stream, for sectors where Australia could hold a genuine comparative advantage in a decarbonising world but where the private sector is unlikely to build the capacity at the pace or scale required. The second is the economic resilience and security stream, for sectors where Australia's exposure to a single supplier or a narrow set of supply routes is judged to create a real national vulnerability, and where paying something above the cheapest available price is treated as insurance rather than waste. The Act also requires the Treasury to conduct and publish sector assessments against those tests, and attaches a set of community benefit principles — conditions about local employment, skills, supply chains and community consultation — to major supported projects (Future Made in Australia Act 2024 (Cth); Treasury, National Interest Framework sector assessments, 2024–25).

That framework deserves to be described accurately before it is criticised. Australia's post-war industry assistance was built on tariffs, quotas and negotiated grants awarded case by case, with no published test and no expiry date. The 2024 Act at least writes the test down, publishes the assessments, and makes the reasoning contestable in public.

The money arrived in stages. The 2024–25 Budget set out a Future Made in Australia package worth roughly $22.7 billion over ten years across its various components (approx.; Australian Government, Budget 2024–25, Budget Paper No. 2). Those components include support for solar manufacturing under the Solar Sunshot program, of the order of $1 billion, and for battery manufacturing under Battery Breakthrough, of the order of half a billion dollars (approx.; Budget 2024–25), alongside expanded lending capacity for the National Reconstruction Fund and the Australian Renewable Energy Agency, and a critical-minerals facility operated through Export Finance Australia. But the two largest single instruments — and the ones that define the agenda's character — are the production tax credits legislated the following year.

The Future Made in Australia (Production Tax Credits and Other Measures) Act 2025 created two of them. The Hydrogen Production Tax Incentive offers A$2 for every kilogram of renewable hydrogen a qualifying project produces, claimable for up to ten years per project, for production occurring in a window running from July 2027 to 2040. Treasury's costing puts the measure at approximately A$6.7 billion over the medium term (Future Made in Australia (Production Tax Credits and Other Measures) Act 2025 (Cth); Australian Government Budget papers and Treasury costings, 2024–25 and 2025–26). The companion Critical Minerals Production Tax Incentive works on the same principle but a different base: rather than a flat rate per unit of output, it offers a refundable offset set at a percentage of eligible processing and refining costs incurred in Australia, for a list of designated critical minerals, over a comparable period and at a broadly similar order of budgeted cost (approx.; same Act and Budget papers). Both are refundable — that is, a company with no tax liability still receives the value in cash — and both are subject to eligibility conditions, including requirements around the timing of a project's final investment decision and compliance with the community benefit principles (approx.; the 2025 Act and its accompanying explanatory materials).

What a production tax credit actually is

The instrument deserves a plain-English definition, because the distinction it draws is the single most important design fact in this entire debate. A grant pays a company money to build something. A production tax credit pays a company money only once something has been produced and sold, at a fixed rate per unit of that output. Under the hydrogen incentive, a company that announces a plant, banks goodwill, holds a ground-breaking ceremony and then never produces a kilogram of hydrogen receives nothing at all. A company that produces a million kilograms receives two million dollars. Nothing is paid in advance; nothing is paid on a promise.

This is not a small point of accounting. Most of the ways governments have historically lost money on industry policy involve paying up-front for capacity that never materialised. A production credit forecloses that failure mode almost entirely, and any fair assessment has to say so before it says anything else.

The evidence

The withdrawal record

Set the policy aside for a moment and look only at what the companies did. Through 2024 and 2025, the pipeline of large renewable-hydrogen projects in Australia — the precise activity the Hydrogen Production Tax Incentive was designed to support — contracted sharply, and it contracted through decisions made by proponents with their own capital and their own boards, not through any change in Australian policy. A Gladstone-region project with an announced capital value of approximately $14 billion was abandoned by its proponent (approx.; proponent announcement, 2025). A Pilbara proposal of an altogether different order of magnitude — an announced capital value in the vicinity of $55 billion — was left in serious doubt (approx.; proponent announcements, 2025). Trafigura cancelled a 440 megawatt South Australian renewable-hydrogen project (Trafigura company announcement). Origin Energy withdrew from the Hunter Valley hydrogen development it had been progressing (Origin Energy company announcement). And the Queensland Government withdrew $1 billion of state support from the Central Queensland Hydrogen Project (Queensland Government announcement).

Two cautions about those numbers, because the honest reading of them is slightly narrower than the arithmetic suggests. First, announced capital values for projects that never reached a final investment decision are not the same thing as capital that was ever going to be spent; the great majority of very large resource proposals never reach construction, in any sector and any decade. Second, the Gladstone project and the Central Queensland Hydrogen Project are both Gladstone-region ventures and the two entries may describe overlapping activity rather than two entirely separate losses. Neither caution changes the direction of the evidence. It changes only how much weight the headline dollar figures can carry, and this article treats them as indicating the scale and breadth of withdrawal rather than as a precise sum of forgone investment.

Why they left: the arithmetic of a kilogram

The withdrawals were not co-ordinated and they were not political. They share a common cause that can be stated in one line of arithmetic. Renewable hydrogen is made by electrolysis — running electricity through water to split it into hydrogen and oxygen — and the electricity is the dominant input cost. Producing a kilogram of hydrogen takes roughly 50 to 55 kilowatt-hours of electricity in a commercial electrolyser, before any losses in compression, storage or conversion for shipping (approx.; standard engineering range reported in CSIRO and International Energy Agency technical assessments). At any firmed renewable electricity price Australia can currently deliver to an industrial site, that alone puts the electricity cost of a kilogram of hydrogen in the range of several dollars, before the electrolyser itself, the desalination plant, the compressors, the storage, the port infrastructure, the finance cost of the whole build, or the additional and substantial energy penalty of converting hydrogen to ammonia for shipping and back again at the other end.

Published cost estimates for delivered renewable hydrogen in Australia have consistently sat well above the cost of the hydrogen the world already buys, which is made from natural gas or coal and sells at a low single-digit dollar figure per kilogram (approx.; CSIRO and IEA hydrogen cost assessments, 2023–25). The gap between those two numbers is the commercial problem, and A$2 per kilogram does not close it. It narrows it. That is a meaningful contribution to a project that is otherwise marginal; it is not a rescue for a project whose costs sit several dollars a kilogram above the price a buyer will pay. And the second half of the problem is the buyer. The export case for Australian renewable hydrogen was always built on Japanese and Korean demand at a price high enough to cover Australian production plus liquefaction or ammonia conversion plus shipping. Firm, bankable, long-dated offtake contracts at those prices have not materialised at the volumes the projects were sized for. A project that cannot sign an offtake contract cannot reach a final investment decision, and a project that cannot reach a final investment decision never gets to the point where a production credit would pay it anything.

This is the point at which it is essential to separate what the Australian government caused from what it inherited and what it has merely presided over. The economics of renewable hydrogen are global. Electrolyser capital costs have fallen more slowly than the industry's own 2020-era forecasts assumed; interest rates rose sharply from 2022, which hits capital-intensive, long-payback projects harder than any other kind; and the Japanese and Korean buyers who were meant to anchor the export trade have moved cautiously. Hydrogen project cancellations through 2024 and 2025 were not an Australian phenomenon — proponents in Europe, North America and the Middle East shelved or slowed flagship projects over the same period. None of that was caused by Australian policy, and it would be dishonest to argue otherwise. Australia's government did not make hydrogen expensive. What it did do was legislate a decade-long subsidy for the activity at close to the moment that global evidence had become clear, and that is a decision about timing and judgement, which is a fair thing to examine.

A subsidy paid on output cannot rescue a project that fails before output — and every project that has walked away has failed at the investment decision, not at the margin.

The other half: why critical minerals is a different argument

Nothing said above about hydrogen transfers automatically to critical minerals, and the distinction is the most important analytical point in this article. The two policies are legislated in the same Act and funded from the same agenda, but they face different markets, different failure modes, and different evidence.

The hydrogen problem is a demand problem: the product costs more to make than buyers will pay, and no amount of Australian policy creates a customer. The critical-minerals problem is close to the opposite. Demand is not in question — battery, magnet and defence supply chains need refined lithium, rare earths, cobalt, graphite and high-purity nickel, and Western governments have made securing them an explicit strategic objective. The market failure, if there is one, sits in price risk and market structure. Australian nickel is the clearest recent illustration. Through 2023 and 2024, a very large increase in Indonesian nickel supply drove global prices to levels at which Australian operations could not cover their costs, and major Australian nickel assets were curtailed or placed on care and maintenance, with the Commonwealth adding nickel to the critical minerals list in response (approx.; company announcements and Commonwealth critical minerals list updates, 2024). The Australian ore did not become worse. The resource did not deplete. A supply decision made elsewhere changed the price, and Australian capacity closed.

That is precisely the risk a private investor must price into a ten-year refinery, and it is precisely the risk that public support can rationally offset — not because governments know better than markets what lithium will be worth, but because the state can bear a category of risk that a single company's balance sheet cannot: the risk that a strategic competitor chooses, for reasons that are not commercial, to make your project uneconomic for three years. Insurance against that is not the same thing as picking a winner. It is closer to the logic of a defence procurement, and it is worth noting that the second stream of the National Interest Framework says exactly that.

The upshot is that a reader who concludes "Future Made in Australia is failing" from the hydrogen evidence has drawn a conclusion the evidence does not support. The correct conclusion is narrower and more useful: one commodity bet inside the agenda is going badly, for reasons that are largely global, and the strategic case for the other is materially stronger and rests on a different and better-evidenced kind of market failure. Any honest audit has to hold both of those at once.

The critique

The knowledge problem

The first objection is not that governments are foolish. It is that they are working with the same information as everyone else, later, and with weaker feedback. A company that misjudges a commodity loses money and stops. A government that misjudges one has a legislated program, a departmental division, a stakeholder list and a set of announcements to defend, and it finds out it was wrong through a process that runs years slower.

The hydrogen incentive is a case study in the timing risk that creates. The Australian hydrogen strategy was framed between roughly 2019 and 2022, when governments, banks and consultancies alike published broadly similar forecasts of falling electrolyser costs and rising Asian demand. Those forecasts were not stupid, and the government was not alone in believing them. But by the time the enabling legislation passed in 2025, a large volume of contrary evidence had accumulated, on the public record, in the form of proponents cancelling exactly the projects the incentive was designed to reward. The legislation went ahead anyway, with a subsidy window running to 2040. The problem is not that a government made a forecast that turned out to be wrong. It is that the policy machinery kept moving after the forecast had been contradicted by the people best placed to test it.

There is a specific design failure inside that. As set out above, the binding constraint on Australian renewable hydrogen is not the operating margin of a running plant; it is the ability to reach a final investment decision, which requires bankable offtake at a price no buyer has yet offered. A production tax credit is paid after production. It improves the operating economics of a plant that exists. It does very little for the financing case of a plant that does not, because a lender assessing a fifteen-year debt facility discounts a credit that only starts once the plant is built, only runs to 2040 regardless of when the plant starts, and depends on future parliaments not amending it. The instrument, in other words, is well designed against fraud and waste and poorly matched to the actual thing that is stopping these projects.

The political-economy problem

The second objection is about what happens next, and it does not depend on the hydrogen bet failing. Industry subsidies are easy to start and hard to stop, because they create a constituency whose livelihood depends on their continuation and whose members are concentrated, organised and located in identifiable electorates, while the cost is spread thinly across every taxpayer and shows up in no one's household budget as a separate line. Australia knows this pattern well: unwinding the tariff wall built around Australian manufacturing took roughly three decades of politically expensive decisions from the 1970s onwards, and the automotive assistance that replaced part of it survived governments of both parties before it ended.

A 2040 end date on the statute book is real discipline, and this article has already given the government credit for it. But statutory end dates are amendable by the parliament that inherits them, and the political economy of extending a subsidy is far more favourable than the political economy of letting it lapse. If a small number of hydrogen projects do reach production on the strength of the credit, they will arrive in the late 2020s with a workforce, a regional footprint and a legislated expiry approaching — which is the exact configuration in which extensions get granted. The risk is not that $6.7 billion is spent. It is that $6.7 billion establishes a claim.

The danger of a ten-year subsidy is not the ten years. It is the constituency the ten years creates, and the parliament that finds it easier to extend than to explain.

What happens to the $6.7 billion

Ask the direct question: if the renewable-hydrogen export sector does not materialise, what happens to the money? The answer is the design's best feature and its most awkward implication at once: because the credit pays only on delivered kilograms, a sector that never produces claims almost nothing. The $6.7 billion is a budgeted provision, not a cheque already written; if nothing is produced, most of it is never paid, and the fiscal cost of the failure is small.

That is genuinely reassuring about the taxpayer's exposure and genuinely unflattering about the policy's purpose. A subsidy whose main protection against failure is that failure costs nothing is a subsidy that has not been asked to do much work. And the fiscal number understates the real cost in two ways. First, a budgeted $6.7 billion occupies fiscal space while it sits on the forward estimates: it is money the Commonwealth has notionally committed and therefore cannot commit twice, at a time when the Commonwealth's own budget papers project structural deficits across the medium term. Second, and more importantly for the people in this article's opening paragraphs, the cost of a government publicly underwriting an industry is not only the money. It is the signal. Announcements move workers between cities, move training providers into new qualifications, move councils into rezoning and move contractors into capital they would not otherwise have spent. Kirsty Waldron's relocation, her short course and her employer's extra apprentices are not on any budget line, and no production tax credit will ever reimburse them. When the state lends its credibility to a commercial proposition, it transfers some of the risk to people who cannot diversify.

The transparency gap

A final, narrower criticism concerns visibility. A grant program produces a list of recipients and amounts. A refundable production tax credit is administered through the tax system, where taxpayer confidentiality provisions apply, and the amounts claimed by individual firms are far less visible than a grant would be. Aggregate cost will appear in the budget papers and the tax expenditures statement, and the Australian National Audit Office retains its capacity to examine the program's administration. But the fine-grained public accountability of "which company received how much, and did it deliver what it promised" is weaker under a tax credit than under a grant. The trade-off runs against the instrument's other virtues: the same mechanism that prevents money going to projects that never produce also makes the money that does flow harder for the public to follow.

What they got right, and the honest caveats

Four things deserve to be said clearly on the government's side of the ledger, and they are not throat-clearing.

The instrument choice is genuinely better than the alternatives it displaced. Of all the ways to lose public money on industry policy, paying up-front for capacity that never appears is the most common and the most expensive, and the production tax credit structurally prevents it. Compare the counterfactual directly: had the same $6.7 billion been deployed as construction grants to the projects announced in 2022 and 2023, a substantial share of it would now be gone, spent on early works for developments that were subsequently abandoned by their own proponents. It is not. That is not luck. It is the design doing its job, and the government chose that design before it knew it would be needed.

The framework is an improvement on how Australia used to do this. A published statutory test, Treasury sector assessments open to public argument, two clearly distinguished rationales, community benefit conditions and a legislated end date together constitute a far more disciplined architecture than the tariff-and-grant apparatus that preceded it. Reasonable people can attack the test's content. It is much harder to argue the country was better off when there was no test at all.

The strategic diagnosis on critical minerals is substantially correct, and being wrong about one commodity does not make it wrong. Australia does hold a genuine endowment; the midstream is genuinely concentrated in a single jurisdiction; that concentration has genuinely been used to make Western capacity uneconomic; and the resulting risk is genuinely of a kind private capital prices at a level that prevents investment. A government that identified that correctly and then chose the wrong horse in an adjacent race has made an error of commodity selection, not an error of strategy. Those should not be scored the same way, and this article does not score them the same way. The parallel companion analysis of the real cost of the energy transition makes the same distinction from the other direction: a correct diagnosis and a costly delivery record can coexist.

And the causes of the hydrogen retreat are overwhelmingly external. Electrolyser cost curves, global interest rates, Japanese and Korean offtake appetite, and the willingness of Chinese producers to accept low margins in nickel and lithium are not Australian policy variables. This government inherited a hydrogen strategy begun under its predecessor, presided over a global repricing it did not cause, and made one decision that is properly its own: to legislate the subsidy anyway, on the timetable it chose. That last decision is the fair target of criticism. The first two are not.

The honest caveat on the other side is that this article is being written before the policy has had a chance to fail or succeed. No renewable hydrogen has yet been produced under the incentive, because the window does not open until July 2027. It is possible that electrolyser costs fall faster over the next three years than they have over the last five, that a Japanese or Korean buyer signs at a price that clears, and that a smaller domestic-use project — hydrogen for ammonia fertiliser or for green iron rather than for export — turns out to be the viable path nobody was modelling in 2022. This article does not claim to know that will not happen. It claims something narrower and better supported: that on the evidence available as at 2025, the private sector's own revealed judgement about export hydrogen has been consistently negative, that the public subsidy runs against that judgement rather than with it, and that a policy which requires the market to be wrong in order to work is carrying a heavier burden of proof than it has so far discharged.